The worst-case secret of the upcoming 7nm Ryzen launch is that these chips will stop with just eight cores. AMD has never come right out and said that it would introduce a 16-core desktop CPU for its Ryzen desktop platform, but it showed off a Ryzen chip with a distinct spot where a second eight-core chiplet could go, and winkingly told us we’d have to wait and see.
Now, noted leaker TUM_APISAK has published new information claiming to know what the clocks are on this upcoming 16-core CPU.
Zen2 ES 16 Core Base clock 3.3 Ghzฺ
Boost clock 4.2 Ghz MB X570
This CPU name can't decode by decode chart PS(ภาพหน้าจออาจจะอัพโหลดให้ในภายหลัง)
The PS means that “Screenshots may be uploaded later.”
All of the usual grains of salt should be applied. This is an ES chip, so it may not be clocked as high as a final product. It’s a leak, which means it could just be wrong (though Apisak’s track record is considered good).
One mark in favor of its accuracy is the significantly lower clocks compared with the incorrect product stack that surfaced last December.
The original 16-core rumors around third-generation Ryzen suggested that AMD would launch 16-core CPUs with a base clock of 3.9GHz and 4.3GHz, respectively, with boost frequencies of 4.7GHz and 5.1GHz, while somehow maintaining very low TDPs (125W and 135W, compared with the modern 2950X’s TDP of 180W).
These clocks and TDPs never made much sense. CPU power consumption is not a linear line; it bends sharply upwards at the top of the frequency range, as voltage increases. The boost clocks weren’t so much the intrinsic issue — AMD could always define a very high single-core boost clock that practically never triggers, much as Intel did with the Core i7-8086K. Modern OS’s are almost always using at least two threads, after all. But the rumored base clocks were very high as well, especially the 16-core chip with a 4.3GHz base.
The fact that this CPU is built for Ryzen’s AM4, not Threadripper’s TR4, will almost certainly impact its capabilities and clocks. We know that 7nm gives AMD substantial headroom, in terms of CPU power consumption — that’s one reason the company can drop a 16-core chip into a socket that launched with an eight-core CPU in the first place. But when it comes to socket loading, the question of whether the motherboard can handle the power demands of the processor isn’t just about TDP. Individual pins within the socket have to be capable of safely handling their own electrical loads.
Ten years ago, Intel’s LGA 1156 socket proved to be prone to failure under extreme overclocking due to the use of Foxconn sockets that didn’t make perfect contact with the CPU’s pads as well as the fact that far more current was being pushed over the tiny connections than they were typically asked to handle. When the people blowing up their equipment are overclockers, AMD and Intel both have some room to shrug, saying: “Try not blowing up your equipment next time.”
But when it comes to a product you are launching yourself, the very last thing you’d want is to build a chip that overwhelmed and damaged X370 or B350 motherboards due to a marginal mismatch between CPU electrical requirements and the underlying capabilities of the motherboard. It’s far better to play things a little conservatively.
This is doubly true when you consider that the 16-core chip AMD is launching into this segment isn’t really going to be the part it leads with in the mass market. Few people make practical use of 16 CPU cores. This is not the company’s mass-market Ryzen or even its enthusiast Ryzen. This 16-core CPU (that’s still technically hypothetical) is AMD’s unparalleled upgrade opportunity for the handful of users who bought Ryzen as a serious workstation and loves the idea of doubling core count on an older system without buying a completely new box. The flip side to that is that the 16-core third-gen desktop Ryzen might not try to push the envelope right to the wall on both core count and clock speed. It makes far more sense for AMD to play the situation conservatively and make certain every chip is an easy drop-in replacement with no underlying concerns than to risk consumer motherboards failing under a 16-core load.
There have also been questions about whether the 16-core version of Threadripper goes away if 16 cores are now possible on the mainstream Ryzen platform. My guess is that it won’t. Threadripper will still have more PCIe lanes, double the memory channels, and a considerably more robust socket to play with. I have no inside information on what Threadripper parts AMD will formally bring to market when that hardware launches*, but it wouldn’t be crazy for the company to deploy a strong 16-core CPU on desktop, in a configuration it knew wouldn’t cause electrical problems for any motherboard, and a more aggressively-clocked CPU on the TR4 platform, with higher performance, more RAM bandwidth, more PCIe lanes, and higher power consumption.
Of course, this is just an ES CPU, so clocks could still tick upwards. AMD has kept a tight lid on exactly what to expect in terms of speeds and feeds. But even if this chip’s clocks represent the final values for a 16-core chip, Ryzen 1 users will be in for a particular treat. Even if both chips maintain a similar top-end all-core clock, third-generation Ryzen CPUs will feature additional IPC improvements to benefit from, and twice the cores on tap. We would expect a CPU with these kinds of specs to be 1.8 – 2x faster than a Ryzen 7 1800X — and if that happens, it’ll be a hell of an upgrade.
*I am aware of fears that AMD has canceled Threadripper after it showed the third-generation parts coming later this year in a March slide deck, but removed any mention from Threadripper in its most recent quarterly slides.